Effectrode LA-1A tear-down and trace adventure

[Shaggy45]

Excellent. I can't wait for you to get this up and running.

[vigilante397]

Seriously though Paul, if you have a minute I would love to hear your thoughts I respect your experience a lot more than my own.

[Rob Strand]

Okay, here's my complete updated schematic. I removed the things I don't plan on using, though I left the taps on the V3 cathode resistors as I haven't decided how I want to switch them yet. Please take a look and let me know if there's anything that doesn't look right

The opamps are marked TL072 but the power is to MCP6021.  Since it's only 5V TL072 should be MCP6021.

I'm assuming 5V power rails is correct not 12V?   5V certainly limits the range of current in the opto (assuming R71 = 5k6 is correct).

D9 is likely to be a zener, perhaps around 2V + 1.5V = 3.5V.   You would need to get that right since it determines the opto resistance with no signal.   3.6V would just cut-off.

[vigilante397]

The opamps are marked TL072 but the power is to MCP6021.  Since it's only 5V TL072 should be MCP6021.

Yup, clerical error on my part. I used the TL072 footprint from my library and just renamed it, but when I re-did the sidechain I forgot to rename it again. It is absolutely MCP6021, there are no other opamps used. Need to update that.

I'm assuming 5V power rails is correct not 12V?   5V certainly limits the range of current in the opto (assuming R71 = 5k6 is correct).

That's correct, MCP6021 has a max input voltage of 5.5V. R71 = 5k6 is correct, but notice they're using 2N7002 as a switch for the 12V rail, the output from the sidechain goes to the gate of the FET, not directly into the LED.

D9 is likely to be a zener, perhaps around 2V + 1.5V = 3.5V.   You would need to get that right since it determines the opto resistance with no signal.   3.6V would just cut-off.

Agreed, based on my measurement in circuit I'm pretty confident it's a 2.8V zener.

[Rob Strand]

That's correct, MCP6021 has a max input voltage of 5.5V. R71 = 5k6 is correct, but notice they're using 2N7002 as a switch for the 12V rail, the output from the sidechain goes to the gate of the FET, not directly into the LED.

Yep, I did notice that.  Actually  when I clicked on your big pic I could easily make out the opamps and R71.

So if the opamp swings to 5V and the MOSFET Vgs_off is about 2V then the output on the drain is about 3V.   I didn't check the LED specs on the optos but say at low current the LED is 1.2V that only leaves 5 - 2 - 1.2  = 1.8V across 5.6k, so 0.32mA max.   Quite low!

Agreed, based on my measurement in circuit I'm pretty confident it's a 2.8V zener.

I found these ones:
ZMC2.4 or ZMC2.7
SOD-106  glass (2mm long), black cathode ring

https://static.rapidonline.com/pdf/47-3629_v1.pdf

They have nominal voltage at 5mA.  The Effectrode is pushing about 10mA into them which will raise the measured voltage somewhat.  Hard to know if it would be the 2.4V or the 2.7V, but in a test circuit it should be clear which one.

[vigilante397]

So if the opamp swings to 5V and the MOSFET Vgs_off is about 2V then the output on the drain is about 3V.   I didn't check the LED specs on the optos but say at low current the LED is 1.2V that only leaves 5 - 2 - 1.2  = 1.8V across 5.6k, so 0.32mA max.   Quite low!

I'm having trouble following your math there. Effectrode is using the MOSFET there as a switch, so when the opamp swings past the threshold voltage (Vgs) the internal body diode is bypassed allowing current to flow from drain to source. Rds_on isn't incredibly low (about 2 ohms), but is low enough that there will only be 100mV or so drop in Vds, so the source of the MOSFET will see pretty close to 12V every time the opamp passes the threshold, which is plenty to light an LED with the 5k6 resistor.

I found these ones:
ZMC2.4 or ZMC2.7
SOD-106  glass (2mm long), black cathode ring

https://static.rapidonline.com/pdf/47-3629_v1.pdf

They are nominal voltage at 5mA.  The Effectrode is pushing about 10mA into them which will raise the measured voltage somewhat.  Hard to know if would be the 2.4V or the 2.7V, but in a test circuit it should be clear which one.

They definitely look the part, to be on the safe side I'll probably order a few different values in case my assumption was wrong.

[Rob Strand]

I'm having trouble following your math there. Effectrode is using the MOSFET there as a switch, so when the opamp swings past the threshold voltage (Vgs) the internal body diode is bypassed allowing current to flow from drain to source. Rds_on isn't incredibly low (about 2 ohms), but is low enough that there will only be 100mV or so drop in Vds, so the source of the MOSFET will see pretty close to 12V every time the opamp passes the threshold, which is plenty to light an LED with the 5k6 resistor.

I'm seeing it as a source follower and not a switch.   That's why I  lose Vgs_off =2V on the source relative to the gate at 5V.   
Maybe you can check because it's easy to screw-up without the board in front of me and I can't see all the tracks.

- 2N7002  =  *N-channel*.    So gate needs to be Vgs above source to turn on.
- Source going to the 5.6k resistor (R71).                     Source = MOSFET pin 2
- Drain going to the plane, presumably that's +12V.   Drain = MOSFET pin 3
- Gate going to the 10k resistor (R70),                          Gate = MOSFET pin 1

So my "wiring" matches your schematic but the MOSFET pin outs are numbered differently.

Numbering as per,
https://assets.nexperia.com/documents/data-sheet/2N7002.pdf

[merlinb]

Tube side-chain is magical thinking. Why not add a unicorn?

Because Effectrode market themselves as true-blue real-mccoy cork-sniffing valve-loving hot-bottle audiophile aristocrats. The LA1A is supposed to be a "high-end, studio-grade vacuum tube compressor in a pedal", based on 'inspired by' the LA2A. So it just deflates me a bit to discover it's an opamp-based optical compressor, with a token tube added for window dressing and no recognisable kinship with the LA2A.

[skyled]

It seems to me like it's a tube preamp with a compressor tacked on.

[Rob Strand]

It seems to me like it's a tube preamp with a compressor tacked on.

At the end of the day all compressors have some sort of "preamp" so a tube compressor could only be expected to have a tube preamp.

[Ben N]

It seems to me like it's a tube preamp with a compressor tacked on.

The signal path is all tube, so that may be just a tiny bit uncharitable, but it's true that the bits that actually make it a compressor are SS, and an LDR is, after all, just an LDR.

[vigilante397]

I've been playing with a couple different options for the layout, and what I really wanted was to put it into a 1590BB, but I was having trouble making it fit the way I wanted. But then I thought if it had one less tube it would be way easier. V1 and V2 are in parallel, so I decided to see how different it sounds with just 2 stages in parallel instead of 4.

Turns out they knew what they were doing. I used a looper and recorded some clips so I could easily do an A/B comparison, and 4 parallel stages just sounds better at every single setting. 2 stages doesn't sound bad and is still an absolutely usable compressor, but 4 stages just sounds better. 1590XX it is I guess

[Ben N]

... so I decided to see how different it sounds with just 2 stages in parallel instead of 4.

Turns out they knew what they were doing. I used a looper and recorded some clips so I could easily do an A/B comparison, and 4 parallel stages just sounds better at every single setting. 2 stages doesn't sound bad and is still an absolutely usable compressor, but 4 stages just sounds better.

Why is that do you think? Superlow Z-out? If that's it, then followers ought to do the trick. Cumulative Miller capacitance? Some sort of cancellation between the parallel paths? Voodoo?

[vigilante397]

Why is that do you think? Superlow Z-out? If that's it, then followers ought to do the trick. Cumulative Miller capacitance? Some sort of cancellation between the parallel paths? Voodoo?

Probably less to do with voodoo and more to do with Miller capacitance and output impedance, if I were to take a guess

[BJF]

Hi There

Parallelling amplifiers is a common technique used in many circuits sometimes with different design objectives: for instance in HF circuits with tubes two triodes can be parallelled to increase useful bandwidth, while in poweramps transistors or tubes are paralleled to increase available current and in line drivers even OP-amps can be parallelled to increase drive capability i.e available current; then there's paralleling devices to minimize noise: There was an RIAA phono preamp made in the 80's with the then and probably today too lowest noise figure transistor ever made the Rohm 2SA1038 and that used three of those PNP's in parallell as a preamp to NE5534A at an impedance corresponding to the source resistans that would yield lowest input noise on an NE5534A≈6K Ohm. It was very impressive 40 years ago while the parts and the application is perhaps history today.

One could wonder why parallelling three low noise transistors used as emitterfollower?
The design objective would obviously be to lower noise and if the final device is to be used infront of a heavily distorted amplifier one can argue that every dB that can be won in Signal/Noise is valueable.

As a story from the 80's some of the local really heavy bands wanted to use wha infront of their roaring Marshalls that were modified for a magnitude of gain and I recall resorting to 2SC2389 for the voltage amplifier and a Bourns plastic trace potentiometer and complementary treble lift ahead of circuit and treble decrease after circuit
18dB ratio at 2KHz both filters-really complicating things was that they wanted a vocal sound requiring a bit of gain
Though a standard wah contains also an emitterfollower there was no significant gain in changing also that to super low noise device but it was left an ordinary NPN.

For a buffer in an application like this ( to be followed by extreme gain) one can choose a discrete transistor and typically a low noise J Fet or Bipolar would have less noise than an ordinary OP amp but also a gain less than 1. In such systems buffers contribute little to noise but anything making voltage amplification does.

Even so if there is pcb room one could certainly parallell three low noise transitors to gain something very small while chances are it will be masked by later stages having high voltage gain


Sidetrack:
Once I needed to make a on off booster that was to be a gift for the 100th of a model and two circuits I considered was one using parallell super low noise bipolars but in the end a low noise J Fet had better overload carachteristics-this required a J fet screened for low noise
and ideally then current should have been choosen for lowest noise but I needed the gm at lower current so there was tradeoff but this circuit was supposed to go after distortion and then overload is a more urgent consideration than absolute lowest noise.

Anyway have fun
BJ

BJF Electronics
Sweden

[vigilante397]

Hi There

Parallelling amplifiers is a common technique used in many circuits sometimes with different design objectives: for instance in HF circuits with tubes two triodes can be parallelled to increase useful bandwidth, while in poweramps transistors or tubes are paralleled to increase available current and in line drivers even OP-amps can be parallelled to increase drive capability i.e available current; then there's paralleling devices to minimize noise: There was an RIAA phono preamp made in the 80's with the then and probably today too lowest noise figure transistor ever made the Rohm 2SA1038 and that used three of those PNP's in parallell as a preamp to NE5534A at an impedance corresponding to the source resistans that would yield lowest input noise on an NE5534A≈6K Ohm. It was very impressive 40 years ago while the parts and the application is perhaps history today.

One could wonder why parallelling three low noise transistors used as emitterfollower?
The design objective would obviously be to lower noise and if the final device is to be used infront of a heavily distorted amplifier one can argue that every dB that can be won in Signal/Noise is valueable.

As a story from the 80's some of the local really heavy bands wanted to use wha infront of their roaring Marshalls that were modified for a magnitude of gain and I recall resorting to 2SC2389 for the voltage amplifier and a Bourns plastic trace potentiometer and complementary treble lift ahead of circuit and treble decrease after circuit
18dB ratio at 2KHz both filters-really complicating things was that they wanted a vocal sound requiring a bit of gain
Though a standard wah contains also an emitterfollower there was no significant gain in changing also that to super low noise device but it was left an ordinary NPN.

For a buffer in an application like this ( to be followed by extreme gain) one can choose a discrete transistor and typically a low noise J Fet or Bipolar would have less noise than an ordinary OP amp but also a gain less than 1. In such systems buffers contribute little to noise but anything making voltage amplification does.

Even so if there is pcb room one could certainly parallell three low noise transitors to gain something very small while chances are it will be masked by later stages having high voltage gain


Sidetrack:
Once I needed to make a on off booster that was to be a gift for the 100th of a model and two circuits I considered was one using parallell super low noise bipolars but in the end a low noise J Fet had better overload carachteristics-this required a J fet screened for low noise
and ideally then current should have been choosen for lowest noise but I needed the gm at lower current so there was tradeoff but this circuit was supposed to go after distortion and then overload is a more urgent consideration than absolute lowest noise.

Anyway have fun
BJ

BJF Electronics
Sweden

Thanks for the insight, that all makes sense

I guess in the matter of a quick update, the PCBs for my first prototype of the compressor came yesterday, I'm still waiting on a Mouser order before I can get it put together. I asked some input from my target audience (bassists) and a few people recommended putting in a gain reduction indicator, so I took a stab at that.

[vigilante397]

Alright, I finished the first prototype tonight and I'm thrilled to report that it is working! The only issue is that I think the output level is a little low, but I'm not A/B-ing it with the original so I'm not positive, I'll double check tomorrow, and if so that's an easy fix. The gain reduction indicator (the 10-bar LED on the front) is working really well, which is awesome because I had never used it before. I need to tweak the values in its threshold as it only goes up 6 bars right now in full compression, but other than that it works great. I'm beyond excited for this thing, hoping to do some more thorough tests tomorrow as it's after midnight here and I have a work meeting in 8 hours. Anyway here's the board right now. You may notice a couple hand-soldered 0ohm resistors, those disconnect the gain reduction indicator; I tested it without first to make sure it was working, then added the resistors by hand to test the whole thing.

[vigilante397]

Well I figured out where the low output level is coming from, and it looks like I have some homework (and probably asking for help ) to do. I have a B+ of 220V, and with V1 and V2 in parallel I have essentially 25k as the equivalent plate resistor (4x 100k in parallel). But for whatever reason the original LA-1A is measuring 136V on the plate but mine is dropping to 87V. The line itself isn't sagging at all, V3 sill gets 220V on the plates. I removed the coupling cap to make sure nothing downstream is messing things up, but no change.

Any thoughts anyone?

[marcelomd]

Nice PCBs! So sleek. I love how the the tube sockets turned out.

Sadly, I cannot help with the circuit =\

[anotherjim]

If all the extra voltage drop is across the plate resistor then one or more triodes are conducting too much? What are the cathodes & grids? Tried tube swap?
Ultra neat PCB BTW!